Anthropic faces lawsuit over use of Elvis Presley music in AI training

The legal challenge against Anthropic

Anthropic is facing a new legal battle after owners of Elvis Presley's intellectual property rights filed a lawsuit on Friday. The plaintiffs claim the company trained its Claude AI models on the music of the rock icon without obtaining any permission or licenses. This move comes as the industry continues to grapple with how large language models ingest massive amounts of copyrighted data.

According to the complaint, the company allegedly used 274 specific songs from Presley's discography to develop its technology. These works include famous tracks like "Can't Help Falling in Love," "Don't Be Cruel," "Return to Sender," and "(You're the) Devil in Disguise." The lawsuit argues that these songs were integrated into the AI's training set without the consent of the rights holders.

The plaintiffs are seeking a permanent injunction to stop Anthropic from using Presley's music in the future. Additionally, they are requesting damages of up to $150,000 for every single work they believe was infringed. Because there are 274 works listed, the total requested fine could reach as high as $41.1 million. Anthropic has stated it intends to defend against these claims fully.

Allegations of copyright infringement

The core of the lawsuit rests on the idea that Claude's generated content now adapts or copies Presley's music in a way that violates copyright law. The complaint alleges that the company established its commercial framework by violating intellectual property rights on a massive level. The plaintiffs argue that the AI is not just learning from the music but is actively reproducing it for users.

The filing describes specific instances where users can ask Claude to recreate or update Presley's existing songs. The plaintiffs claim these results often maintain the same structure and hooks while only using light rewording for the verses and choruses. They argue that this level of similarity moves past mere inspiration and into the territory of reproducing copyrighted material.

The legal team argues that even when the specific words are swapped out, the underlying structure of the song remains a violation. They contend that the AI is essentially providing a way for users to generate new versions of protected works. This creates a situation where the AI's output could potentially replace the original music in various commercial or creative contexts.

A pattern of legal disputes

This lawsuit is not an isolated incident for Anthropic, which has faced several other challenges regarding intellectual property. In August, Sony Music Publishing and Warner Chappell Music sued the company for a "brazen campaign" of illegally torrenting and scraping copyrighted works. These companies expressed concerns about the massive scale at which data is being downloaded to train AI models.

The company has also dealt with significant legal settlements in the past involving the use of protected content. In July, Anthropic settled a $1.5 billion class action lawsuit where authors accused the company of illegally downloading pirated books. These events suggest a broader trend of creators seeking to protect their work from being used to train competing technologies.

Furthermore, Universal Music Group has filed two separate lawsuits against Anthropic for copyright infringement. One of these was filed in January, while the other was initiated in October 2023. These recurring legal actions indicate that the music and publishing industries are actively monitoring and challenging the data practices of major AI developers.

The impact on AI development

For the AI industry, these lawsuits represent a significant hurdle in determining the legality of training data. If the court finds that Anthropic infringed on Presley's rights, it could force companies to change how they gather information. Developers might have to seek individual licenses for every song or book used in their training sets, which could significantly increase costs.

The current situation puts pressure on the business models of companies like Anthropic. If they are forced to remove specific copyrighted works, it could affect the capabilities of their models. However, the company maintains that it disagrees with the claims and will defend its practices in court to protect its current operations.

For the average user, the outcome of this case could determine how much of the cultural history they can interact with through AI. If the court rules in favor of the plaintiffs, certain types of creative requests might be restricted or removed from the Claude platform. The balance between fair use and copyright protection remains one of the most contested areas in modern technology law.